I previously raised a question about abnormal tagging on PRoW Oct-14. This is an expanded version of the same query including additional information. Public Rights of Way in eastern Hampshire (UK) are being tagged as access=private (the mapper states this is because they have been established by easement and the landowner has a right to prevent access). This results in ways losing their identity as footpaths or bridleways, being indistinguishable from genuine private ways on rendering used by OSM Standard, OSM Cycle Map & OSMAND. Routing for hikers, cyclists and other leisure users on OSMAND is wrecked, the routing software avoids footways, bridleways & tracks tagged access=private, even though low level tags of foot=yes etc have been used. The accepted usage for PRoW has always been access=yes, followed by low level tags where these are needed to indicate barred forms of traffic. This form is recognised throughout the Wiki, with access=private reserved for genuinely private roads & paths where access is conditional on the land-owner's assent. Reversing this creates substantial problems for software developers and creates a whole new set of exceptions that need to be programmed in. In (UK) legal terms the situation is clear. Landowners do not have any right to prevent public access to a registered PRoW for the designated class of traffic. Landowners do not have any right to stop up or abandon a PRoW. The route by which a PRoW was established has no bearing on established access rights and any attempt to thwart access is a criminal offence. Most rural PRoW pass over private land (either established by usage or agreement). If mappers accept reversed tagging then it is logical to extend this to all PRoW over all ground where there is title. It is also logical to extend it to roads that have been built under PFI contracts, where the facility is owned by a private company, so the Severn Crossings, M6 Toll etc can all be tagged as access=private (with sub tag motor_vehicle=yes). OSM is an international map, that implies mappers have some responsibility to maintain discipline in tagging, following established practice whenever possible. If mappers take highly individualistic views of how tags are applied we have mapping anarchy and the consequence of that is OSM will collapse under its own contradictions, leaving the market to the high cost producers who will always create a 100% reliable product. Unless this question is resolved I can have no confidence in OSM data used in routing applications and that loss of market confidence will spread out to more and more leisure users in the UK if they find their hiking or cycling route planning impossible using OSM data. asked 16 Feb '15, 16:36 Sailor Steve SK53 ♦
showing 5 of 7
show 2 more comments
|
The question has been closed for the following reason "This is a discussion not a Q&A topic." by SK53 23 Feb '15, 00:10
OK - I'll try and provide some sort of answer to the several issues here. Whilst I suspect that it's really one for discussion on the talk-gb mailing list, I'll try and provide some answers to what I suspect are the key questions (and the answers to your previous question are also worth reading). Question: Is it right or wrong to add " I'd argue that it would be irrelevant (and daft) to include a general access tag on a Question: Is it right or wrong to add " Here's where it gets interesting. If I see a " If I see " However, it's quite common for designated public footpaths to run along private farm tracks. Often there'll be a "private" sign to accompany the "public footpath" one. Here I'd suggest that an " Question: What about Public Bridleways? They're similar to footpaths, except that the " Question: ... and Restricted Byways? Essentially as "public bridleways" but also allowing non-motorised vehicular traffic, so if you fancy a trip out in the coach and four, you're legally allowed to do so down a " Question: ... and a Byway Open to All Traffic? The clue's in the name - this implies " Question: What does It means "whatever the sort of traffic that would normally allowed down here are". On a Question: What does It probably means that the previous mapper was confused. I've always interpreted it as " Question: What about footpaths, bridleways and tracks that aren't designated as public footpaths, bridleways or byways, etc.? Use access tags as best you can to describe the access rules that you know to be in force, based on your knowledge of the area. This might mean any one of a myriad of possible combinations - try and pick the best one based on what you see. Question: Do any of the 5 map styles available on OpenStreetMap.org show England and Wales style footpath designations and access rights? No. Question: Would it make any sense of one of them did? As things stand, not really - the osm.org styles all have to make sense internationally, and an English and Welsh public footpath rendering wouldn't even make sense across the whole of the UK. Question: Would it be nice if someone ponied up the cash for some web hosting to host a tileserver that did show this? Yes - feel free to contribute and organise. It'd have to be as a "local add-on" though. Question: Do any "off the shelf" routers understand England and Wales style footpath designations? Not that I'm aware of, though sanity in access tagging (saying "foot=yes" on footpaths and public tracks, omitting "access=private" when it is not needed etc.) will help greatly. Question: How do I make my Garmin car satnav avoid dirt tracks that I don't want it to route down? I find that removing them from the data before running mkgmap works for me. It does mean that you can't then use it for foot routing, but that's usually not an issue with a car satnav. Question: I have a disagreement with another mapper. What's the best way to resolve it? I'd start by trying to discuss it with them, and trying to understand each other's point of view. If that doesn't work, try the local mailing list (or forum, or facebook page, or whatever else your locality uses). If that doesn't work, try the OSMF Data Working Group(5). It's unlikely that the DWG will be able to wave a magic wand, but they'll at least be able to offer a view independent of the disagreeing parties. Notes Firstly - all of the above are my views, based on a fair bit of mapping of public rights of way in the last 7 years or so. There are different approaches taken (see for example SK53's answer to your previous question - a quick chat on talk-gb or #osm-gb will reveal that there are no certainties here). If there are takeaway messages from any of this it's "don't assume there are any certainties; alway do an on-the-ground survey of rights of way; always talk to other mappers to try and appreciate their point of view". (1) bicycle is an interesting one here - there are people who'll argue that you have a right to cycle on public footpaths, but let's leave that to one side for now. (2) in the absence of anything else preventing it, such as a traffic regulation order. (3) I don't believe I can think of a single exception. (4) in England and Wales - other countries have far more sane access rules. (5) full disclosure - that currently includes me. answered 16 Feb '15, 19:31 SomeoneElse ♦ 1
meta (e.g. @SomeoneElse): conversion into a comment does not work for me here (as it happens often) - apparently (if I saw correctly) for you too. meta @Sailor Steve: could you please add this as a comment under the answer to which it belongs to ("add new comment" link)?
(16 Feb '15, 23:37)
aseerel4c26 ♦
1
Just to add my take on this. Firstly, the designation tag is the primary tag to define whether something is a right of way. However I'm not convinced about access=private for a public right of way. If there's an explicit Private sign along the ROW then maybe, yes it makes sense, but I don't consider the mass tagging of rights of way in Hampshire with access=private, whether the path is marked "Private" or not, helpful at all. My inclination (if I get the time) would be to remove all these access=private tags in Hampshire, but will not do so if the consensus is that these tags are correct.
(18 Feb '15, 13:45)
nickw
@nickw I described
(18 Feb '15, 13:52)
SomeoneElse ♦
|
Thanks for your very detailed summary of tagging issues affecting PRoW & paths in general. I have seen suggestions in questions that there could be an intermediate tag created that sits between access= yes/private perhaps access=conditional followed by sub tags detailing the permitted access modes which might help here and be internationally useful. I will take the whole problem to the DWG to see if they can add clarity. answered 16 Feb '15, 23:19 Sailor Steve |
My understanding of access= is that it's a shorthand way of setting all the relevant modal access rights to the same value. You can then use individual modal keys to over-ride that default setting. access= is not supposed to be a summary of the overall access rights to a way. So "access=private, foot=yes" implies a route that private to all modes apart from pedestrians, and it's the correct was of recording such rights. Personally, when I'm tagging PRoWs, I usually only add the relevant positive modal values. So foot=yes for a Public Footpath, foot=yes, bicycle=yes, horse=yes for a Bridleway, etc. (Actually, I'll use designated in place of yes a lot of the time -- but that's an unnecessary complication for the present discussion.) I don't think there would be anything wrong with adding an access=private in addition to these tags, if the route is indeed private for all other relevant modes. But I do think that adding the individual modal tags is necessary -- we don't really want to force routers and outer data users to have to know about the legal PRoW background in England and Wales, and know that designation=public_footpath etc implies some particular access tags. If each country did something like this, it would place an unreasonable burden on international projects. Also, if we were to do that, we'd then have the problem of whether or not the implied access tags over-ride ones that are given explicitly. e.g. is "designation=public_footpath, access=private" open to pedestrians and is "designation=byway_open_to_all_traffic, motor_vehicle=no" open to motor vehicles? answered 18 Feb '15, 13:38 Robert Whitt... |
Disclaimer: It is my work that is being discussed in this question. I also consider it impolite that I was not invited to join the larger community discussion especially since I was happy to engage with Steve via PM. "the mapper states this is because they have been established by easement and the landowner has a right to prevent access" That is not strictly true. My point was that a public footpath only grants a right of way to pedestrians, not other forms of traffic. Following the conventions of OSM as described in the wiki access=private, foot=yes is the correct way to comprehensively describe which modes of transport can access a public footpath (and just as importantly which ones cannot). Doesn't designation=public_footpath imply foot=yes? I think it is unfair for us to assume that data consumers worldwide will be familiar with the intricacies of English rights of way. After all nobody would suggest that we should omit admin_level=10 from admin boundaries because it is implied by designation=civil_parish. It would also be necessary for any data consumer to check that the way in question is within the boundaries of England and Wales which might be a complex and costly operation. "This results in ways losing their identity as footpaths or bridleways, being indistinguishable from genuine private ways on rendering used by OSM Standard, OSM Cycle Map & OSMAND" The answer to this is to fix the data consumers rather than trying to remove valid data from the database. Displaying access details on the standard OSM rendering has always been a problem because we cater for many different transport types often with conflicting access permissions. Strictly speaking the current rendering is correct, provided you are a motorist. answered 19 Feb '15, 23:26 AndyS Andy - I looked for your user name before making my post, but as it is not the same as your mapping name this led to a dead end. On our PM exchanges I did say I was going to take the issue to the community, you could have given me your OSM-Help details.
(22 Feb '15, 21:56)
Sailor Steve
1
this answer really highlights that this is a topic for a mailing list discussion & not the Q&A format for which this site is designed. I am now closing it for that reason, PLEASE move further discussion to talk-gb.
(23 Feb '15, 00:09)
SK53 ♦
|
I have a strong interest in this as cycle.travel carries out more detailed analysis of access tags than, I believe, any other end-user website. I agree that it isn't reasonable for worldwide sites to parse E+W-specific values of the designation= tag (to my knowledge, cycle.travel is the only one that does). However, it is widely understood and accepted that
This is why we moved away from tagging horse=yes on every single road back in 2006 (you can look it up in the lists if you like ;) ). The international default is that highway=footway is restricted to pedestrians. This is also the case for 99% of country-specific values. A few places allow bikes under some circumstances and Brazil appears to allow mopeds. That's it. Consequently it is clearly documented, and universally accepted, that there is no need to tag access=private on a footway to prevent motor vehicle access. So these tags are unnecessary. A second question is whether they are wrong. In a technical sense, no, they aren't, for the reasons described by @AndyS. In a community sense, however, intentionally breaking the visualisation used by the most popular renderers is antisocial at best and vandalism at worst. I would therefore expect that these tags should be removed and that such action, should it prove necessary (which I hope it won't), would be backed up by DWG. As a postscript, @Sailor Steve, I think you would probably get a bit more sympathy if you could make your comments a little shorter and save us the moralising about "OSM will collapse under its own contradictions" and "loss of market confidence", which frankly just makes you look pompous. answered 20 Feb '15, 10:36 Richard ♦ I'd disagree, defaults are suggested values to data consumers rather than a guarantee from the mapper. According to the "UK tagging guidelines" highway=footway is unpaved but head into any town and you'll find plenty that are missing surface=asphalt tag because the mapper did not consider the surface when mapping. There is also the no entry in that wiki page for highway=track. What are the "default permissions" for a track? It is somewhat perplexing to see that you consider that it is unreasonable for a consumer to parse E+W versions of one tag but not another.
(20 Feb '15, 11:53)
AndyS
I also find it unhelpful that you suggest I am "intentionally breaking the visualisation" - I am not. My aim is to accurately and explicitly record which transport modes are permitted to use a particular route.
(20 Feb '15, 11:54)
AndyS
2
@AndyS - if you don't add "access=private" to a "highway=footway", do you really think that someone could reasonably interpret the resulting data as "I could drive a car down there"?
(20 Feb '15, 11:56)
SomeoneElse ♦
@SomeoneElse - That question is somewhat misleading as the access tag is a legal issue not a physical one. A better question might be "Can a motorcycle go down a path?" or "Can a car go down a track?".
(20 Feb '15, 12:09)
AndyS
1
@AndyS I covered "private" tracks in my answer below. I specifically said "access=private" on "highway=footway" because that's what's in question here - the use of an "access" tag that adds no value to the data but does break OSM Carto's rendering (a sort of "tagging against the renderer", if you will). Another issue is replacing "foot=yes" or "foot=designated" with "access:foot=yes" - whilst a human could figure that out, it will confuse routers as it's not an obvious tag.
(20 Feb '15, 12:15)
SomeoneElse ♦
@SomeoneElse - What we are dealing with are mostly paths/tracks not footways but yes I consider an explicit declaration to be useful for the same reasons as surface=unpaved on a footway is useful. Making explicit declarations of "default" values shows that you have considered the property in question and wish to record that fact. As a data consumer I'd have more faith that a footway was unpaved if it had a surface tag rather than relying on a prescribed default value. The same principal applies to access and other tags.
(20 Feb '15, 12:58)
AndyS
@SomeoneElse - Regarding "access:foot" etc. I no longer tag these to the exclusion of the common "foot" style tags so this is no longer an issue. I'd be happy to discuss my use of those tags with you in a more appropriate venue.
(20 Feb '15, 12:59)
AndyS
showing 5 of 7
show 2 more comments
|
Just to pick up on one point - you've stated that "The accepted usage for PRoW has always been access=yes, followed by low level tags where these are needed to indicate barred forms of traffic.". Would it be possible to link to places where this was previously discussed and accepted?
The answers you got last time when you asked essentially this question were (paraphrasing) that "designation=public_footpath implies foot=yes", so the rest of the tags are arguably irrelevant to specifying foot access.
(and one other comment that must be made I think)
"Reversing this creates substantial problems for software developers and creates a whole new set of exceptions that need to be programmed in."
No, I don't believe that it does.
I regularly use a map style that renders access based on designation in England and Wales. It's actually very easy to do:
https://github.com/SomeoneElseOSM/designation-style
SomeoneElse:
I referred to these links before making my post - https://help.openstreetmap.org/questions/4737/best-practices-for-applying-accessprivate also came up in this one - https://help.openstreetmap.org/questions/35408/rights-of-way-mapping-united-kingdom
I have made a visual check across southern England looking for other instances of tagging PRoW as private. I could not find any other than where PRoW cross or follow private driveways (and I think this situation has also been the subject of discussion). I do not know if there is any way of interrogating the who UK data set for particular tag combinations.
Throughout my working life (45 years or so) I used maps on a daily basis - to me as a user I want accuracy and consistency of product. Your link to a map style is really no use to me or most users without technical knowledge. OSM products must be 'ready to use' for hikers, bikers, riders motorists et al without having to struggle through forums & obscure links to find work-arounds for structural failures. OSM cannot be allowed to degenerate into anarchy otherwise it will collapse. The community model works most of the time, but it has limits and community leaders need to take this on board.
Re "OSM products must be 'ready to use' for hikers" they are - I use OSM data every day. I don't have to "struggle through forums & obscure links"; I just look at the web browser on my phone or at the screen of a Garmin GPS and there it is.
It sounds like you have a problem with a particular rendering (which is not designed to show rights of way in England and Wales - as an "international" style it would make no sense to) and you'd like mappers to adapt their tag usage so that it looks "there or thereabouts" on this map style?
Is there a question here, or is this really one for the lists?
The rendering on OSM Standard & OSMAND show PRoW with an access=private tag as private ways. This means map users are unable to distinguish between genuine private paths/tracks and PRoW without recourse to an Ordnance Survey map. I agree that the England & Wales tail should not wag the worldwide dog, to achieve this all that is needed is to ensure PRoW are not tagged as private.
OSM may have started out here, but I'm not convinced that an England-and-Wales specific rendering really belongs as one of the OSM "standard" ones. It'd be nice if you could pick your own tile layers rather than resorting to a cludge like http://wiki.openstreetmap.org/wiki/User:SomeoneElse/Your_tiles_from_osm.org , but given that everyone developing and running the site is a volunteer, it's understandable that every feature that everyone might want can't be delivered yesterday (we got routing today - yay!).
On the second point, I suspect that maps showing access based on "designation" in OsmAnd would be feasible. I've not looked at OsmAndMapCreator, but if the worst came to the worst you could process the pbf file before running OsmAndMapCreator on it.